Gridded Simulation Study Of Monthly Precipitation Distribution In China Under Terrain Of Different Scales Supported By BEMD Terrain Decomposition

Topography has an extremely significant and complex influence on the spatial and temporal distribution of precipitation.It is one of the main factors leading to the local weather anomaly in the weather system.The distribution of terrain in China is complex,small terrain and medium and large scale terrain are intricate.Part of the small terrain can enhance the precipitation distribution,and part of the small terrain has the same effect on the precipitation distribution as noise,so removing this part of the small terrain can help to improve the accuracy of precipitation fitting.However,due to the wide distribution of small terrain in large and medium scale terrain,small terrain can not be quantitatively defined or simply removed.In this paper,the idea of Bidimensional Empirical Mode Decomposition(BEMD)is creatively introduced.By gradually extracting the micro terrain,the small terrain is smoothed,and then the terrain with different scales(different degree of smoothness)is constructed.Then we analyze the influence of DEM elevation and aspect on the spatial distribution of precipitation,and prove that there are differences in the spatial distribution of precipitation under different scale terrain.Finally,we establish a monthly precipitation model based on different scale terrain to verify the above results and analyze the influence of small terrain on precipitation.The main work of this paper includes the following aspects.(1)This paper decomposes the DEM data of the whole country.The DEM with spatial resolution of 1km is used as input data,and the two-dimensional empirical mode decomposition(BEMD)algorithm is introduced to solve the key problem of terrain decomposition.The DEM Terrain Data is decomposed,and the 8-level micro terrain with frequency from high to low and its corresponding residual function are obtained.(2)Based on the results of terrain decomposition,combined with the landform classification index,the landform classification is realized.Based on the results of terrain decomposition,the concept of micro terrain wavelength is established.With the help of the relationship between wavelength and frequency,micro terrain frequency is classified:Micro terrain is divided into high frequency micro terrain,medium frequency micro terrain and low frequency micro terrain,and then the macro residual terrain is divided into high frequency residual terrain(orig3 scale terrain)Medium frequency residual terrain(orig5 scale terrain)and low frequency residual terrain(orig8 scale terrain).Based on the micro topography and macro residual topography,combined with the indexes of altitude and topographic relief,the geomorphology can be divided into three major categories and 22 sub categories.(3)The difference of the influence of different scale terrain on the spatial distribution of precipitation is analyzed.When the terrain scale changes,the DEM elevation,slope and aspect of the terrain will change,which will lead to the change of spatial distribution of precipitation.Among them,the influence of aspect change on the difference of precipitation distribution is the most significant.The change of aspect has the most significant effect on the precipitation distribution.With the increase of terrain scale,the influence of local terrain on the spatial distribution of the prevailing precipitation direction gradually decreases;The fitting accuracy of the spatial distribution of the prevailing precipitation direction(the consistency between the prevailing precipitation direction and the prevailing wind direction)showed a trend of first increasing and then decreasing;In different geomorphic areas,the change of the prevailing precipitation direction is inconsistent.The change of aspect makes the distribution of aspect factor change:the distribution of windward and leeward slope of mountain is different under different scale terrain;With the increase of terrain scale,the fitting degree of aspect factor first increases and then decreases;The terrain scale with the highest fitting degree of aspect factor is affected by the landform type.The change of terrain scale also leads to the change of precipitation difference between the main and the secondary precipitation directions.The precipitation difference of main and secondary precipitation directions increases with the increase of terrain scale.The precipitation difference between the main and the secondary precipitation directions is related to the monthly precipitation.The larger the precipitation is,the greater the precipitation difference between the main and the secondary precipitation directions is.In orig3?orig5 terrain,the change rate of precipitation difference between the the main and the secondary precipitation directions is the largest,which indicates that the extraction of medium frequency micro terrain has a greater impact on precipitation concentration,while the extraction of high frequency and low frequency micro terrain has a smaller impact on precipitation difference.(4)A grid monthly precipitation model based on different scale terrain is built.The accuracy of the model under different scales is compared by building comprehensive evaluation index.The optimal model is selected and the influence of small terrain on precipitation distribution is analyzed.Based on the TRMM Precipitation Data,the monthly precipitation estimation model B₀-B₃(B₀corresponds to DEM scale terrain,B₁-B₃corresponds to orig3,orig5 and orig8 scale terrain)is built.The correlation coefficient,average absolute error and average relative error are selected as precision indexes to construct comprehensive evaluation index evaluation models.With the increase of terrain scale,the accuracy of the model first increases and then decreases,and reaches the maximum in model B₂.Compared with TRMM retrieved precipitation model A,the MRE and MAE of precipitation model B₀with terrain factor decreased by 4.54%and 1.72%,respectively,indicating that terrain has a significant impact on precipitation distribution.The MRE of model B₂is 1.13%lower than that of model B₀,and MAE is 0.77%lower than that of model B₀,indicating that the medium and high frequency small terrain has interference effect on precipitation fitting.